random.

It skewed better when tests became more available. Early on, you were given a test if you were symptomatic so you likely had it: P( covid | symptoms of covid). Then, tests became more available so the pool was less likely to have it regardless of what was actually happening in the population. it was P( covid | want a test). Obviously P(covid | sympoms of covid) >>>> P( covid | want a test)

I can imagine (though I certainly can't quantify it) at this moment its skewing worse as some of that pent up desire to get a somewhat pointless test to feel good has gone down.

The skewing doesn't tell us that things are better or worse than the metric shows, it shows that we need to work a little harder to figure out what's happening.
[Post edited by jmanatVT at 12/11/2020 12:37PM]